format: decode selected chunks into reusable buffers

A selection read (a hyperslab or points covering at most half the dataset) decoded each chunk it overlaps into fresh buffers, one per filter stage; it now uses the thread's chunk-decoding scratch like the full readers. Covered by tests/chunked_read_paths_interop.rs (small hyperslabs and points over every filter and type) and the partial-read equivalence tests. Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
2026-09-26 10:20:07 -05:00
parent 9e9b849dd7
commit 9e608b975c
2 changed files with 52 additions and 49 deletions
@@ -31,7 +31,7 @@ use crate::lane_partition::PartitionStats;
 /// [`DecodeScratch`]), kept between reads so decoding reuses memory instead
 /// of faulting in fresh pages for every chunk. A re-entrant call (a
 /// registered filter codec that itself reads a file) gets a fresh scratch.
-fn with_scratch<R>(f: impl FnOnce(&mut DecodeScratch) -> R) -> R {
+pub(crate) fn with_scratch<R>(f: impl FnOnce(&mut DecodeScratch) -> R) -> R {
    #[cfg(feature = "std")]
    {
        use std::cell::RefCell;